Crashes on roadways continue to represent a major global public health concern due to high rates of death and injury, underscoring the need for predictive tools that can identify high-risk conditions and guide prevention strategies. This study develops a framework that combines structured crash records and road information with unstructured police narratives to predict injury severity using machine learning and natural language processing (NLP). The dataset is used to train, validate, and test nine models, combining three algorithms (Random Forest, AdaBoost, and XGBoost) with two NLP methods (TF-IDF and Word2Vec). Model performance is evaluated using macro-average F1-scores to address severe class imbalance. Results show that XGBoost with TF-IDF achieves the best performance (macro-F1 = 0.644), demonstrating measurable improvements from incorporating narrative features compared to structured data alone. Beyond prediction, a simulation-based sensitivity analysis is conducted on the top 100 features, identifying 11 variables with the greatest impact on severity outcomes in Kentucky. Seatbelt non-use, occupant entrapment, and impaired driver control emerge as the most influential factors, with simulated improvements leading to notable reductions in fatalities and major injuries. The study introduces a “prediction-to-prevention” framework that links injury severity prediction with simulation-based sensitivity analysis. By integrating structured and narrative crash data, the framework identifies how changes in key behavioral and roadway factors can shift injury outcomes toward less severe levels. These findings highlight the dual contribution of this study: improving predictive accuracy through narrative integration and offering actionable insights to support evidence-based traffic safety interventions.
1 March 2013 in Journal of Transportation Engineering
Horizontal curves present drivers with numerous tasks that, if not performed while approaching and negotiating the curve, may result in a roadway departure crash. A vehicle’s lateral position within the lane and its speed are two indicators of interest from safety and operational perspectives. These can and have been measured simultaneously at multiple locations along the curve. However, researchers face the challenge of collecting operational data while minimizing impacts on driver behavior, and developing robust, efficient, and accurate means to obtain the data. This paper presents the findings of a series of pilot studies on closed and open courses that investigated the effectiveness and accuracy of pneumatic road tubes and digital video cameras for collecting such data. Closed-course studies investigated a single data collection station setup, whereas an open-course study investigated multiple data collection stations on a horizontal curve. The data were reduced manually and automatically, and...
1 August 2016 in Journal of Transportation Engineering
Precise milepoint measurement data are essential for better subway track management and maintenance practices within railroads and subways. For milepoint estimation, dead reckoning systems of some subway track geometry cars use as pesudolite markers having no unique identification information. In such cases, milepoint measurement data have to be postprocessed. However, the postprocessing is conducted in a manual fashion and is time consuming and labor intensive. This paper presents an optimal match method to automatically postprocess milepoint measurement data. The presented method consists of three submodels: (1) dynamic-programming-based distribution-pattern match model for differentiating actual markers from false-positive ones, (2) correlation-analysis-based algorithm determining milepoints for recognized markers, and (3) a linear interpolation equation for milepoint revision. The method was applied to 124 inspection runs for 15 tracks of the Beijing subway system whose track geometry car is s...